Bull. Mater. Sci., Vol. 9, No.3, August 1987, pp. 207-217.
Crystallization behaviour of MetgIas
©
282~
Printed in India.
MB (Fe40Ni3SMo4BlS)
V SRAJA, KISHORE and S RANGANATHAN
Department of Metallurgy, Indian Institute of Science, Bangalore 560 012, India
MS received 21 May 1987
Abstract. Differential scanning calorimetry, x-ray diffraction and transmission electron
microscopy have been employed to. determine the thermal stability and identify the
crystalline phases on devitrification of Metglas 2826 MB. The glass crystallizes into y..
FeNiMo and fcc (FeNih3 B6 with activation energies of 270 and 375 k} mol - 1 respectively.
The reactions are primary and polymorphic in nature. The influence of Mo towards
crystallization of Fe 4 0 Ni 4 0 B2 0 has been to/enhance the formation of the fcc (FeNih3 B 6
phase in preference to orthorhombic (FeNihB phase and to raise the thermal stability of the
amorphous state.
Keywords. Crystallization; Kissinger plot; microstructure; metallic glass.
1. Introduction
The crystallization study of metallic glasses is of interest as it provides a means to
understand the nucleation and growth processes in undercooled melts anddetermine
the thermal stability of glasses (Ranganathan and Banerjee 1984; Ranganathan and
Suryanarayana 1985). It also derives interest from the fact that a number of
metastable phases can be synthesized through this route. We have studied the
thermal and chemical stability of a series of Fe-Ni-B glasses and this study of
Metglas 2826 MB forms a part of it (Raja 1986).
Metglas 2826 MB (Fe4oNi3SMo4Bls), a product of Allied Corporation (USA),
has drawn considerable interest towards understanding its thermal stability
(Antonione et a11978, 1981; Azam et al1979; Mizgalski et a11981; Nicolai et al1981;
Battezzati et al 1982; Cumbrera et al 1982a, b; Menon and Suryanarayana 1985).
Employing non-isothermal DSC measurements, Antonione et al (1978) reported
two-stage crystallization processes at 703 K and 777 K having activation energies of
296 and 334 kJ mol- 1 respectively for the first and second crystallization reactions.
However, Majumdar and Nigam (1980) and Cumbrera et al (1982b) observed a
three-stage crystallization of the glass. DTA measurement by Majumdar and Nigam
(1980) showed crystallization peaks at 728, 810 and 863 K. Cumbrera et al (1982b)
determined activation energies both from Kissinger (288 and 355 kJ mol- 1 ) and
isothermal (288 and 451 kJmol- 1 ) methods using DTA, DSCand resistivity
measurements for the first two crystallization processes. It may be noted that the
activation energy deduced from the isothermal method for the second crystallization
differs significantly from that obtained by the Kissinger method.
Mizgalski et al (1981) studied the sequence of crystallization using XRD. Their
studies indicate that the first phase to crystallize from the glass is the fcc (FeNi)23 B 6 ,
followed by the formation of a y-FeNiMo solid solution. At high temperature
(FeNil.Bcbct phase was produced. An orthorhombic (FeNi)3B phase was found in
samples annealed in a vacuum of 2 x 10- 5 torr. Interestingly, Battezzati et al (198.2)
reported the presence of an orthorhombic Ni 3B phase in Metglas 2826 MB annealed
at about 1000K even without subjecting the glass to vacuum.
207
-
.
','
208
V S Raja, Kishore and S Ranganathan
The observations by Nicolai et al (1981) are interesting in this context. They
interpreted their DSC and Guinier high temperature camera experiments to
indicate several stages of reaction. Phase separation was observed as a precrystallization process. The first crystal to form was y-FeNiMo solid solution. This was
followed by the formation of fcc (FeNi)23 B6 phase by a peritectoid reaction of the yphase with the remaining glass of composition (FeNiMo)7s B2s. Eventhough
reactions other than primary, eutectic and polymorphic have been speculated upon
in the literature (Ranganathan and Banerjee 1984), this is the first time that a
peritectoid reaction has been claimed to occur.
In a limited but direct microstructural study by TEM, Cumbrera et al (1982a)
observed the appearance of y-FeNiMo followed by fcc (FeNi)23B6' Subsequent
heating to high temperature led to a stable phase whose structure was not
determined.
Menon and Suryanarayana (1985) have recently brought out the difference
between the crystallization behaviour of the vapour deposited and liquid quenched
Metglas 2826 MB using TEM. The vapour-deposited glass leads to a fcc solid
solution, which is different from the y-FeNiMo phase formed during annealing.
Furthermore; it has been pointed out that the vapour-deposited glass yields a simple'
cubic phase instead of the fcc boride phase crystallizing in melt quenched glass.
As seen above, the review of the literature on Metglas 2826 MB shows that, in
spite of extensive work, diverse views exist among several investigators regarding the
crystallization process. The present study is undertaken to address the following
points over which differences exist in the literature: (a) the number of stages of
crystallization (two, three or more); (b) the order of appearance of the crystalline
phase (y-FeNiMo and (FeNiMo)23B6)); (c) the nature of the second stage of
crystallization (Polymorphic or peritectoid); (d) the possibility of phase separation
preceding crystallization.
2. Experimental procedure
Metglas 2826 MB obtained from Allied Corporation, USA was in the form of
ribbons of width 15 mm and thickness 35 usi: The nominalcomposition in atomic
percentage was Fe40Ni3SMo4B1S. A Perkin Elmer DSC 2C was used to determine
the crystallization temperatures. In' order to get a true value of crystallization
temperature free from the thermal gradients of the instrument, the method of Meisel
and Cote (1983) was adopted. The phases present in the annealed samples were
analysed using a Philips 1140 x-ray diffractometer (XRD) with CoK« radiation of
wavelength 0·17902 nm. The microstructures were observed under, a Philips 301
electron microscope operating at 100 kV. Thin samples for transmission electron
microscopy were obtained using a Tenupol twin jet electropolisher controlled by
Struers variable potential stabilizer. A good polishing was obtained with an
electrolyte of 20% methanol and 80% sulphuric acid maintained at -10 c e and with
an operating, DC volt of 10.
3. Results and discussion
3.1
Differential scanning calorimetry
The glass subjected to constant heating rates in DSC for 10,20,40 and 80 K min- 1
iP
-210
V S Raja, Kishore and S Ranganathan
9·0
~o..
e
-
lGO
11·0
1·4
1·2
Figure 2. Kissinger plots for the first and second crystallization process.
Table 1. Comparison of activation energies for the crystallization of Metglas 2826 MH.
MSI·
kJ mol- 1
kJ mol"!
MSII
270
296
*403
288
288
375
334
415
355
451
Method of
determination
References
Kissinger
Ozawa
Kissinger
Kissinger
Isothermal kinetics
Present work
Antonione et al (1978)
Nicolai et.al (1981)
Cumbrera et al (1982b)
"
*The values correspond to the second and third peaks, where y-FeNiMo and fcc(FeNih3 B6
are reported, whereas these phases have been found to crystallize under first and second
peaks of the DSC curve according to other investigators.
identified by heating the glass till the end of the peaks in DSC, cooling it back to
room temperature with a cooling rate of 320 K min -1 and subsequently analysing it
in XRD. Phase transition in the crystallized glass, if any, at high temperature is
identified by heating the glass in DSC upto 1000K. These XRD results are
summarized in figure 3. The XRD of the amorphous alloy is extremely broad (figure
3a). The k value (= 4n sin e/2) determined from the half width of the half maxima
turns out to be 31·1 nm - 1. The broad peak becomes sharper in the case of the
sample annealed till the end of the first peak of the DSC trace (figure 3b). The peak
corresponds to (111) of y-FeNiMo phase with the lattice parameter 0·358 nm. The
other peaks are not visible owing to the low volume ofy-FeNiMo present in the
sample. Distinct peaks appear in the x-ray diffractogram (figure 3c) of the glass
annealed till the end of the second exothermic peak which are indexed satisfactorily
on the basis of fcc (FeNi)23 B6 phase with a = 1·061 nm. It should be noted that the
peak corresponding to (111) of y-FeNiMo phase merges with that of (333) of
(FeNi)23B6 in this case. The glass annealed upto 1000 K shows significant volume of
y-FeNiMo phase which might have transformedfrom fcc (FeNi)23 B 6 boride phase.
5
Crystallization behaviour of Metglas 2826 MB
m
Q)
<.)
u
(;;
an
u:;
0)
u
Q)
s
U
....,
N
IX)
u '>-
- ::!
211
r;;
C'l")
0)
~
0
N
N
en
0
\0N
-...:I'
0"'-3'
0_
N
>
!:::
(/')
Z
....UJ
=
(b)
(0)
30
1.0
50
DEGREES, 2&( Co Kc~J
Figure 3. X-ray diffractograms showing the sequence of crystallization: (a) amorphous
(b) heated upto 760 K (c) heated upto 860 K and (d) heated upto 1000 K. (C.B. refers to
cubic boride).
3.3
Transmission electron microscopy
The microstructure and the.corresponding selected area diffraction (SAD) pattern of
the as-received amorphous alloy is shown are figure 4. The glass possesses a mottled
structure and the unusually broad halo rings (especially the first ring) complement
the XRD of the amorphous alloy shown in figure 3a. This may be attributed to phase
separation. Such an instance of phase separation in an as-quenched glass has earlier
been reported in Fe-B alloys by Walter and Bartram (1978). However, the mottled
structure may be due to the preferential dissolution during electropolishing.
Cumbrera et al (1982a) have indeed explained their microstructure on this basis. As
has been observed by Walter and Bartram (1978), two sets of haloes are not detected
in this glass. Further, there was no evidence for two glass transition temperatures.
Thus incipient phase separation remains only a possibility, without adequate
evidence. It has to be pointed out that among the many investigators, only Nicolai et
al (1981) have postulated phase separation.
The development of microstructure has been studied for the isothermal annealing
at 773 K using TEM. Figure 5 shows the microstructure, SAD pattern and dark
field image (corresponding to the second ring of the pattern) for the glass annealed
for 15 min at 773 K. The SAD pattern can be indexed satisfactorily for y-FeNiMo
phase with a=O'358 nm and the dark field image reveals almost spherical crystals
;
Crystallization behaviour of Metglas 2826 MB
209
between 320 and 1000 K exhibited two exothermic peaks (the typicalDSC trace for
the heating rate 10K min - 1 is shown in figure 1). This agrees with the results of
Antonione et al (1978) but shows disagreement with other results which reported
three (Majumdar and Nigam 1980; Cumbrera et al 1982b) and four (Nicolai et al
1981) exothermic peaks. However, it can be noted that the peak temperatures for the
first and second peaks reported by these investigators agree well with the present
values. Additional support for the two-stage crystallization is found in the fact that
for a variety of (FeNiMo)B glasses, Antonione et al (1980) have found only two
stages. The nature of crystalline phases formed under these peaks will be discussed in
§§ 3·2 and 3·3 using XRD and TEM.
The peak shift with heating rates has been exploited to determine the activation
energy for crystallization employing Kissinger's (1957) equation
In PIT; =
-EeIRTp+C,
(1)
where P, T p , E; and C respectively represent the heating rate, peak temperature,
apparent activation energy and a constant. The Kissinger plots of In PIT; vs IITpare
shown in figure 2. The E; values turn out to be 270 and 375 kJ mol- 1 respectively for
the first and second crystallization processes. The values compared with earlier
results reported in the literature are given in table 1. In a majority of the cases the
present values agree with the earlier reported results. However, two differences can
be noticed: (i) Nicolai et al (1981) reported higher values of activation energies for
both the transitions and (ii) Cumbrera et al (1982b) estimated a higher value from
the isothermal kinetics for the second crystallization reaction. It should be remarked
that the activation energy determined by Cumbrera et al (1982b) using nonisothermal kinetics is in close agreement with the present results. It is also relevant to
note that though the apparent activation energies obtained through DSC and
microscopy techniques have been widely employed to compare the thermal stabilities
of amorphous alloys, Berkenpas et al (1986) expressed reservation towards
meaningful interpretation of the crystallization mechanisms.
3.2 X -raydiffraction
The phases crystallizing out during the first and second exothermic reactions are
o
iex:
UJ
::J:
o
X
w
t
u
~
0:::
W
:I:r--------
8
z
---'
w
920
320
TEMPERATURE K
Figure 1. DSC trace of Metglas 2826 MB for the heating rate of 10K. win -1~
•
212
V S Raja, Kishore and S Ranqanathan
Figure 4.a. Bright field image. b. SAD pattern of the amorphous alloy. Note the
mottled structure in the micrograph.
embedded in the matrix. The fcc (FeNi)23 B6 crystals start emerging from the
specimen annealed for 30 min and their number and size increase with time. A·
complete crystallization is noticed in the alloy annealed for 120 min. Figure 6 shows
the microstructure of the partially crystallized glass (773 K, 45 min). The crystals
possess a faceted morphology. Figure 7 shows the bright field image, SAD pattern
and the dark field image of the fully crystallized (773 K, 120 min) glass. On close
examination some of the fcc (FeNi)23 B6 crystals exhibit twins/faults in them,
Figure 5. a. Bright field image. b. Corresponding SAD pattern. c. Dark field image of
the glass annealed at 773 K for 15 min. The continuous rings belong to y~FeNiMo phase.
>
Crystallization behaviour of Metglas 2826 MB
Figure 5.
213
e
214
V S Raja, Kishore and S Ranganathan
Figure 6'.WeU grown fcc (FeNih3 B 6 crystals in the glass annealed at 773 K for 45 min.
Interestingly, fcc (FeNi)23 B6 crystals grow with the same morphology at low
temperature (723 K, 60 min) as illustrated by the micrograph of figure 8.
The following scheme of sequence of phase transition in Metglas 2826 MB can be
derived from the DSC, XRD and TEM studies.
Amorphous-ey-FeNilvlo + amorphous
(primary crystallization)
~y-FeNiMo + fcc (FeNi)23B6
(polymorphic crystallization)
~y-FeNiMo + fcc (FeNi)23B6
(volume of y-FeNiMo found increasing at the expense
of fcc (FeNi)23B6 phase).
It can be noted that fcc (FeNi)23B6 phase is metastable (Herold and Koster 1978)
and hence can transform into y.. FeNiMo phase. The sequence of formation of yFeNiMo followed by fcc (FeNi)23B6 is in accordance with the observations of
Nicolai et al (1981) and Cumbrera et al (1982a). However, Mizgalski et al (1981)
reported a different sequence as follows
Amorphous-sfcc (FeNi)23B 6 +amorphous
(FeNi)23B 6 +y-FeNiMo-4
fcc (FeNi}23 B 6 +'t:FeNiMo + bct(FeNi)3 B 6 .
~fcc
Comparing the above sequences of crystallization with the present one indicates that
they have not identified the initial formation of y-FeNiMo from the glass but have
reported the phase transition from fcc (FeNi)23B6 to y-(FeNiMo) phase ill the
crystallized glass.
Figure 7.. a. Microstructure. b. SAD pattern. c. Dark field image of the
crystallized glass (annealing treatment 773 K~ 120 min).
fully
Crystallization behaviour of M etqlas 2826 M B
Figure. 7.
215
,>
216
V S Raja, Kishore and S Ranganathan
Figure 8. Microstructure of Metglas 2R26 MB annealed for 60 min at 723 K.
In addition to the y and boride phases, Mizgalski et al (1981) noticed the
formation of bet (FeNi)3 B above 950 K and Battezzati et al (1982) identified an
orthorhombic Ni 3B phase at about 1000 K. However, the present study does not
indicate any of the two phases either in x-ray or in TEM studies. Interestingly,
neither Nicolai et al (1981) nor Menon and Suryanarayana (1985) found the presence
of bet (FeNi)3 B and orthorhombic Ni 3B phases.
The influence of Mo towards the crystallization of FeNiB glasses is more striking
when the crystallization characteristics of Fe 40Ni 40 B 20 glass is compared with that
of Metglas 2826 MB. Raja (1986) has reported that the glass Fe4oNi4oB2o
crystallizes into four types of crystals. The first two of them are eutectic mixtures of
v-FeNi + orthorhombic (FeNi)3 Band y-FeNi + fcc (FeNi)23 B6 and the third and
the fourth are a single phase y-FeNi solid solution and an unknown Fe-rich phase.
Although we find fcc (FeNi)23B6 phase in Fe4oNi4oB2o, its volume fraction is < 5%
(could not be identified by x-ray diffraction and was possible to detect only in TEM).
Hence it is very obvious that Mo enhances the formation of a fcc (FeNi)23 B6 phase
in preference to the orthorhombic (FeNi)3 B phase. Further, in the presence of Mo,
crystallization takes place by simple two-stage processes in contrast to the complex
single-stage crystallization in Mo-free Fe 40Ni 40 B2 0 glass simultaneously imparting a
thermal stability to the glass.
4. Conclusions
(i) XRD and TEM studies have indicated that the glass crystallizes in two stages into
y-FeNiMo followed by fcc (FeNi)23B6.The same sequence is reported by Nicolai et
al (1981) and Cumbrera et al (1982a). Additional stages observed by other
investigators are due to the transformation of the crystalline state to more stable
states.
(ii) The y-FeNiMo phase forms by the primary mode as noted by Nicolai et al
1
.,.
1.
-p
Crystallization behaviour of 1\4 erglas 2826 M B
217
(1981); while the fcc (FeN'i)23B6 forms by a polymorphic mode and not by the
peritectoid mode as suggested by Nicolai et al (1981).
(iii) The activation energies for the devitrification of the glass into y-FeNiMo and fcc
(FeNi)23 B6 phases respectively turn out to be 270 and 375 kJ mol- 1 from the
Kissinger plots.
(iv) Though the appearance of the pseudograin structure and the broad XRD peak
can be attributed to the incipient phase separation of the glass during melt
quenching itself, it is felt that the evidence for phase. separation is not strong.
(v) It is confirmed that the addition of molybdenum enhances the thermal stability of
(FeNiB) glasses and leads to the formation of fcc (FeNi)23 B6 in preference to
orthorhombic (FeNi)3 B.
Acknowledgements
The authors are thankful to the Department of Science and Technology,
Government of India for financial assistance.
References
Antonione C, Battezzati L~, Lucci A. Riontino G and Venturella G 1978 Scr. Metal!. 12 1011
Antonione C, Banezzati L. Lucd A, Riontino (1, Tabasso M C and Venturello (3 1980 J. Phys. (Paris) C41
131
Antonione C, Battezzati L, Lucci A, Riontino G and Venturella G 1981 La M etall. 73 118
Azam N, Lenaour L, Rivera C, Grosjean P, Sacovy P and Delaplace J 1979 J. Nucl. Mater. 83 298
Battezzati L, Antonione C and Cossolo A 1982 Z. A1 etallkd. 73 185
BerkenpasM B, Barnard J A. Ramanujam R V and Aaronson H I 1986 Scr. Metall. 20 323
Cumbrera F L, Millan M, Conde A, Marquez Rand Vigier P 1982a J. Mater. Sci. 17 861
Cumbrera F L, Miranda H, Conde A, Marquez Rand Vigier P 1982b J. Mater. Sci. 17 2677
Herold U and Koster U 1978 Z. tv! etallkd. 69 326
Kissinger H E 1957 Anal. Chern. 29 1702
Majumdar A K and Nigam A K 1980 J. Appl. Phvs. 514218
Meisel L V and Cote P J 1983 Ac( a j\'1 clall. 31 1053
Menon J and Suryanarayana C 1985 Z. A1etallkd. 76 802
Mizgalski K P, Inal () T, Yost P o and Karnowsky M M 1981 J. Muter. Sci. 16 3357
Nicolai H P, Kopmann G and Frommeyer G 1981 Z. Metallkd. 72 558
Raja V S 1986 Thermal and chemical stabilities (~l three Fe-Ni-B metallic glasses, Ph.D. Thesis, Indian
Institute of Science, Bangalore
Ranganathan S and Banerjee S 1984 Meta/lie qlasses-s-production. properties and application (ed.)
T R Anantharainan (Switzerland: Trans-Tech Publications) p 147
Ranganathan Sand Suryanarayana C 1985 Muter. Sci. Forum 4 173
Walter J Land Bartram S F 1978 Rapidly quenched metals (ed.l B Cantor (London: The Metal Society)
Vol. 1, P 307
-:..